Research On Fault Diagnosis Of Beer Fermentation System Based On Improved Principal Component Analysis

The beer production process has a long cycle,many factors affecting product quality,and the fermentation device has a complicated structure,which is a typical nonlinear,strongly coupled,time-varying system.In the actual production process,system failures such as equipment aging,sensor failure,and operating point drift often occur.The occurrence of these faults may cause large economic losses or even personal safety accidents.In this paper,the beer fermentation process is taken as the object to study the data-based intermittent process fault diagnosis method.Based on the research and analysis of various data-driven process monitoring and diagnostic models,the main component analysis fault diagnosis method is improved.Experiments and simulations verify the performance of the improved fault diagnosis algorithm.The main research contents and work are as follows:Firstly,a method for fault diagnosis of beer fermentation system based on independent component analysis is proposed.The independent component analysis optimization criterion is used to decompose independent component components of non-Gaussian variables from the observed variables.It satisfies the independent characteristics in statistical sense,and not only It is only irrelevant that is required by the PCA.Determine the number of independent component components selected by non-Gaussian ordering idea and establish corresponding statistical control confidence limits.Secondly,a method for fault diagnosis of beer fermentation system based on independent component analysis is proposed.The independent component analysis optimization criterion is used to decompose independent component components of non-Gaussian variables from the observed variables.It satisfies the independent characteristics in statistical sense,and not only It is only irrelevant that is required by the PCA.Determine the number of independent component components selected by non-Gaussian ordering idea and establish corresponding statistical control confidence limits.Thirdly,aiming at the nonlinear problem of process data,the principal component analysis method is improved.An improved principal component analysis(KPCA)algorithm based on kernel transformation is proposed,and then the advantages of independent component analysis in data independence analysis are developed.An ICA-KPCA fault detection model,the performance of the algorithm has been significantly improved in terms of missed detection and accuracy.Through the simulation analysis of several fault diagnosis models,it is concluded that the ICA-KPCA method has better performance in the actual false detection rate and missed detection rate compared with the traditional fault diagnosis method.